WO2012042176A1

WO2012042176A1 - Technique for obtaining, via a first node, information relating to path congestion

Info

Publication number: WO2012042176A1
Application number: PCT/FR2011/052273
Authority: WO
Inventors: Fabrice Guillemin; Isabelle Hamchaoui; Thierry Le Maguer
Original assignee: France Telecom
Priority date: 2010-09-30
Filing date: 2011-09-29
Publication date: 2012-04-05
Also published as: FR2965689A1; EP2622802A1; US20130185454A1

Abstract

The invention relates to a method for obtaining, via a first node (10), information relating to congestion of a path that enables packets to be routed from said first node to a second node (14) in a packet communication network (1), said congestion potentially degrading said routing. The method for obtaining information includes the following steps implemented by the second node: a step of receiving (F1) a first packet that is from the first node and intended for the second node, routed via at least one third node (12), said first packet being provided in order to trigger the transmission of a control message in a control plane in which the first and second nodes exchange control messages; a step of processing (F2) said first packet in the control plane, at the end of which a transmission of the control message intended for the first node is requested; a step of obtaining (F3) said information from the first received packet; and a step of sending (F4) a second packet to the first node, said second packet carrying said control message and including said obtained information.

Description

Technique for obtaining, by a first node, information relating to congestion on a road

The invention relates to a technique for obtaining, by a first node, information relating to a congestion of a road between this first node and a second node. More specifically, this route allows routing of packets from the first node to the second node.

The invention is in the field of telecommunications and more particularly in the field of detecting congestion in packet communication networks.

The invention finds a particularly advantageous application for session admission control in such packet communication networks.

Congestion is likely to occur on any equipment located on the path of a package. More specifically, equipment may be congested at input or output memories, internal queues, etc.

Congestion notification functions have been defined by standardization bodies for packet communication networks. These functions are based on a variety of mechanisms, including an explicit ECN congestion notification mechanism for Explicit Congestion Notification, an explicit forward congestion notification mechanism (FECN) for Forward Explicit Congestion Notification. mechanism for explicit backward congestion notification BECN, for "Backward Explicit Congestion Notification" or even combinations of these different mechanisms. The term "before" corresponds to the direction of transmission of a packet, that is to say to the recipient of the packet, and the term "backward" corresponds to the opposite direction of transmission of the packet, that is to say to the sender of the packet. These different mechanisms use a header field of the data packets to carry information relating to the congestion of one of the nodes carrying the packet. Thus, this information relating to congestion is transmitted in the data transfer plan. For example, for an IP-based communication network, for "Internet Protocol", the Internet Engineering Task Force (IETF) document, RFC 3168 specifies the manner in which an IP packet carries an explicit ECN congestion notification in before. Such an IP packet is subsequently called a marked packet.

For the TCP transport protocol, for "Transmission Control Protocol", relying on the IP protocol, it is expected, for a TCP session established between a sending node and a receiving node, that the receiving node notifies the sending node of a TCP segment of the occurrence of congestion when sending an acknowledgment. The sending node can then modify certain parameters of the TCP protocol in an attempt to eliminate the congestion situation. No information on the location of congestion is available in this case. Still for this TCP transport protocol, studies have been conducted to report backward congestion. For example, the June 1998 IETF draft-salim-jhsbnns-ecn-00 document proposes to use control plane messages to notify the occurrence of rearward congestion. Specifically, the ICMP "Source Quench" control message for "Internet Control Message Protocol," defined by RFC 792, is used to notify the packet sender. Such a message notably allows a node carrying a packet to implicitly notify the sender of the packet that it has been deleted or marked by the router. The message includes a part of the deleted package. Various optimizations have been proposed, for example in the article "Congestion Control in TCP / IP networks: a combined ECN and BECN approach" by F. Akujobi et al, published in the proceedings of the MILCOM 2003 conference, in order to limit the sending these ICMP "Source Quench" messages.

However, these solutions have the disadvantage of requiring a significant interaction between the control and transfer plans since the control plane must obtain from the transfer plan the deleted or marked packet. As a result, notification using these messages requires processor and memory resources at the notifying node. These solutions also have the disadvantage of involving a load back link, that is to say the node that sends an ICMP message "Source Quench" to the node transmitting the packet, this load being related to the transmission of these messages. These messages are also likely not to reach the sending node and the latter is not able to detect a loss of one of these messages. The sending node also receives too much information, which it can not use, because of the sending of a message for each deleted or marked packet. In addition, these methods are only applicable for an established TCP session.

One of the aims of the invention is to remedy the shortcomings / disadvantages of the state of the art and / or to make improvements thereto.

According to a first aspect, the subject of the invention is a method for obtaining, by a first node, information relating to a congestion of a route allowing routing of packets from said first node to a second node in a network. packet communication, said congestion being capable of degrading said routing, said obtaining method comprising the following steps implemented by the second node:

a step of receiving a first packet from the first node and to the second node, routed via at least one third node;

a step of processing said first packet in a control plane in which the first and second nodes exchange control messages, after which a transmission of a control message to the first node is requested;

a step of obtaining said information from the first received packet; a step of sending a second packet to the first node, said second packet carrying said control message and comprising said information obtained.

This first packet is intended to trigger the transmission of the control message in the control plane.

In known manner, the control plane allows the different nodes to exchange control information between them. The transfer plan is in charge of routing packets between the nodes.

A route is used to route packets from a first node to a second node, also called a destination node. This road crosses one or more intermediate nodes. The congestion is likely to affect different resources of these intermediate nodes: it can be a congestion affecting a queue of the second node, a congestion affecting hardware resources of the second node, ... The road is located then also congested.

The information relating to congestion in the downstream direction, that is to say in the direction of packet transmission, is determined in the transfer plan by one of the congested intermediate nodes and is intended to mark the packets in this sense. transmission, once routed through the congested intermediate node. According to the invention, the information relating to congestion on the road, received in a first packet, is copied into a second packet carrying a control message transmitted in response to the first packet. This second packet is transmitted to the first node in response to the first packet. Thus, this information relating to congestion of the road is provided solely on solicitation of the first node and is transmitted backwards, that is to say in the opposite direction to the transmission direction of the first packet. The load generated by the implementation of the method according to the invention is reduced both at the level of the hardware resources and processor of the second node at the level of the links taken to reach the first router.

It is emphasized here that the first packet is a packet that is destined for the second node and that requires sending a response control message to the first node.

The information relating to congestion of the road corresponds, for example, to the value taken by the CE bit of the ECN information field of the header of a packet, as specified in the IETF RFC 3168 document. value of "1" indicates congestion.

Only a minimal modification of the operation at the level of the control and / or transfer plans is required, namely the copying of the congestion information in the response packet. The interaction between the control and transfer plans is also not changed. Thus, in case of persistence of the congestion, the notification is made only once whereas in the solutions of the state of the art, a large number of packets is likely to be transmitted. from the transfer plan to the control plane to generate ICMP Source Quench control messages.

Congestion notifications are not sent spontaneously, that is to say in the absence of solicitation, and it is thus possible for the first node to detect a lack of response to the sending of the first packet. The first node can then re-transmit the first packet to the second node. Congestion notifications are also out-of-band, that is, they do not require a TCP session.

This allows fine detection of traffic fluctuations and localization of even evanescent congestion in the network. It is possible to cut the route between the first and second nodes into a plurality of segments, some of which correspond, for example, to the connection between two subnetworks, and to know the state of congestion of each of the segments. On the contrary, the information provided in the TCP acknowledgment messages is global information on the route taken and it does not allow to precisely locate the elements of the network or segments of the road causing the congestion.

The obtaining method also offers an alternative to congestion detection using SNMP procedures for Simple Network Management Protocol. These procedures take into account traffic variations over longer periods of time by the update and export by the nodes of traffic counter values to central collectors according to the SNMP protocol. The reaction time required by these procedures is more important.

Thanks to the obtaining method, it is thus possible for a network operator to know the level of congestion of certain elements of the network or segments by interrogation. This allows you to protect or modify ongoing sessions. This protection can go through the implementation of admission control procedures for new sessions or even rate modulation. In the case of rate modulation, it is possible to use, for example, so-called scalable or scalable SVC coding techniques for Scalable Video Coding. A rate increase is then possible when the load conditions of the network allow it, for example an increase in the available flow or an absence of congestion.

According to a particular characteristic, the first and second nodes implementing differentiated routing of packets according to classes of service, the first packet received comprises information relating to a given class of service and the information obtained relates to congestion affecting the service. least said class of service.

It is then possible to obtain information relating to a congestion of the route for the routing of the packets marked as belonging to the given class of service.

The network operator can then take measures to protect the sessions corresponding to certain classes of service. Alternatively, it is possible for the intermediate node to indicate congestion for all service classes higher than the given class of service. This allows a single request to obtain congestion information for a set of classes of service.

According to another particular characteristic, the control message sent corresponds to a response message to another control message received in the first packet.

It is possible to provide a new control message in a control plane protocol, this new message requiring the sending of another control message in response.

This other control message is an echo request transmitted from the first node to the second node.

Such another control message is for example the "Echo Request" message associated with the "Echo Reply" message for the ICMP protocol.

According to yet another particular feature, the header of the first packet indicates that a lifetime of said first packet has expired.

It is still possible to obtain the congestion information on a path using the utility program, known as "Traceroute", which makes it possible to determine the route taken by a data packet. In this case, the expected response to such a packet corresponds to an error check message. For the ICMP protocol, it is for example the message "Time Exceeded".

According to yet another particular characteristic, the second packet being transmitted to the first node via at least a fourth node, this fourth node does not modify the header of the second packet.

It is emphasized here that the routes between two nodes respectively in the upward and downward directions are not necessarily identical.

When congestion is likely to also occur in the uplink direction, i.e. from the second node to the first node, the intermediate nodes should not change the header of the second packet in the transfer plan. It is thus sufficient to modify the operation of the intermediate nodes in the transfer plan for the control messages used in the implementation of the method of obtaining. The operation of the nodes in the control plane is not modified.

According to a second aspect, the invention relates to a node in a packet communication network, arranged to exchange with another node control information in a control plane,:

receiving means, arranged to receive from the other node a first packet; means for processing said first packet in the control plane, arranged to request a transmission of a control message to the other node at the end of the processing of the first packet;

means for obtaining information relating to a congestion of a route for routing packets from the other node and to said node from the first packet;

sending means (106), arranged to send a second packet to said other node, said second packet carrying said control message and including said obtained information.

According to a third aspect, the invention relates to a communication system in a packet communication network, wherein first and second nodes exchange control information in a control plane, said system comprising:

said first node, comprising sending means arranged to send a first packet, for which the second node must send in response a control message;

said second node being arranged as described above.

It is recalled here that the first packet is provided to trigger the transmission of a control message in the control plane.

According to a fourth aspect, the invention also relates to a computer program comprising instructions for implementing the obtaining method according to the first aspect, implemented by a node of a packet communication network, when this program is executed by a processor.

The invention will be better understood with the aid of the following description of particular embodiments of the method of the invention, with reference to the appended drawings in which:

FIG. 1 is a simplified representation of a packet communication network according to a particular embodiment of the invention;

FIG. 2 represents a simplified diagram of the steps of the method of obtaining and a synoptic of the exchanges between the different entities implementing the method of obtaining according to a particular embodiment of the invention;

Figure 3 is a simplified reminder of the structure of a packet;

FIG. 4 represents a node according to a particular embodiment of the invention. In Figure 1 is shown schematically a network 1 packet communication. This network 1 comprises a plurality of node devices, of which five of them 10, 12, 14, 16, 18 are shown in FIG. 1. Thereafter, these node devices are simply called nodes. These nodes play the role of routers in the network 1, that is to say they are in charge of routing packets in a plane, called transfer plan. A route, also called a path, allows routing of packets from a first node to a second node through other nodes as appropriate. In another plane, called the control plane, the nodes exchange control information, for example using the protocol ICMP. It is understood that the connectivity of a node is not restricted to the simplified network as shown in Figures 1 and 2. A node is indeed able to route packets to a plurality of other nodes.

Subsequently, information relating to a congestion of a road is called information that indicates the existence of congestion for at least one of the intermediate nodes of the road. This congestion is likely to degrade a routing packets along this route in the transfer plan. When this congestion occurs at a node of the road, for example at a queue intended to memorize the packets at the output of the node before retransmission to the next node on the road, the packets are marked by the node before retransmission using an ECN information field, for "Explicit Congestion Notification". This information field includes a CE information subfield, for "Experienced Congestion". The packets are thus marked by setting the information subfield CE to the value one. By way of illustrative example, when the node 14 is congested for the routing of the packets received from the node 12 and to the node 18, through the node 16, these packets are marked at the output of the node 14 and arrive thus marked at the node 18.

According to the invention, when a receiving node is to transmit a control message upon receipt of a first packet from a requesting node, the receiving node obtains the congestion information from the first received packet. and transmits the control message in a second packet, in which it copies this information. This thus enables the requesting node to receive downstream traffic congestion information, i.e. from the requesting node to the receiving node. Thus, the requesting node is able to determine on a route taken by the packets to the receiving node that at least one intermediate node is congested.

Thus, a congestion notification mechanism provided in the transfer plane for a first communication direction is used to obtain information relating to congestion of the road and this information is looped back in a second direction of communication opposite to the first direction. This loopback only occurs for certain first particular packets that will trigger the transmission of given control messages. These first packets are detailed later in connection with FIG.

It is then placed in the case where the nodes 10, 12 and 14 implement the method of obtaining according to a particular embodiment. More specifically, information relating to congestion on the route of the node 10 to the node 14 is necessary at the node 10.

Figure 2 schematically shows the steps of the method of obtaining in this particular embodiment. In a step El, the node 10 transmits a first packet PI to the node 14. In the context of the present description, FIG. 3 recalls the elements constituting the packets routed in the packet communication network 1. The header of the packet consists of a set of information fields from Party to @dst.

In order not to overload the description, the consolidated information fields in the Parti, Part2 forms are not further detailed here, since they are not concerned by the implementation of the obtaining process.

The information field CoS 30 includes information relating to a class of service, when the communication network 1 implements differentiated routing of packets according to classes of service.

The information field 31 corresponds to the ECN information field previously described.

The TTL information field 32 for "Time-To-Live" indicates a number of nodes through which the packet can still be routed. Its value is decremented with each routing by a node. When its value becomes zero, the life of the package has expired.

The information field 33 P makes it possible to determine from which protocol the packet originates.

The information field 34 @src corresponds to the address of the sending node of the packet.

The information field 35 @dst corresponds to the address of the destination node of the packet.

These different fields of information thus constitute the header of the packet.

The packet further comprises an information field 36 which contains the payload data of the packet. For example, it includes an IGMP type control message when the information field 33 indicates the value of two.

This first packet passes through the node 12 in the transfer plane. This is where node 12 encounters congestion in its internal queues for routing packets to node 14. In accordance with RFC 3168, node 12 sets the information subfield CE to the value one in the header of the first packet PI and thus obtains a second packet P2 that it transmits to the node 14.

The second packet P2 is received by the node 14 in a step F1.

In a step F2, the node 14 processes the second received packet P2 in the control plane. At the end of this processing, a control message must be sent to the sending node of the first packet, that is to say the node 10. The node 14 transfers the control message to the transfer plan so that it is routed.

In the transfer plan, the node 14 obtains in a step F3 the information relating to congestion from the second received packet P2. In a step F4, the node 14 constitutes the header of a third packet P3 to the node 10 and copies the information relating to the congestion obtained in step F3. This third packet also includes the control message to be sent.

Still in this step F4, the node 14 transmits the third packet P3 to the node 10. The third packet P3 is routed via one or more intermediate nodes.

In a step E2, the node 10 receives the third packet P3 and has thus obtained information relating to the congestion of the road from itself and to the node 14.

This information is not provided spontaneously or in association with a packet that was to be forwarded down but triggered by a requesting node only.

The obtaining method can be integrated in a more general method, in which the node 10 successively interrogates the different nodes 14, 16 composing the path carrying packets to the node 18. Such a path can be obtained using a routing protocol. It can also be obtained using a utility program, called Traceroute, whose operation is detailed later in connection with a particular embodiment.

The node 10 can then use this information relating to a congestion obtained for the route to the node 18 to admit or not a new session to a destination connected to the node 18. The node 10 can also use this information relating to a congestion to modulate communication rates.

The node 10 can insert first IP packets according to certain rules, for example either regularly in the data stream, or regularly for a set of communications between itself and a set of destinations, for example all the clients connected to a network. same network node, such as a DSLAM multiplexer, for "Digital Subscriber Line Access Multiplexer".

In a first embodiment, the first packet PI carries an ICMP control message "Echo request". In this case, during the processing step F2 of the received packet, the node 14 detects that it must send another control message in response to the ICMP control message "Echo Request". This is an ICMP message "Echo Reply". The third packet P3 carrying the message "Echo Reply" is modified to allow to loop information congestion as received. The changes to be made to the nodes to implement the method of obtaining are thus limited.

In a second embodiment, node 10 implements the previously mentioned Traceroute utility program. This program consists of sending packets indicating a TTL validity period, for "Time-To-Live", increasingly large. Each node retransmitting the packet decrements the validity period. When it becomes zero, the node does not transmits the packet but transmits an ICMP control message "Time exceeded" to node 10. In this second embodiment, during the processing step F2, the node 14 detects that it must issue a control message in response to the received packet. This is an ICMP "Time exceeded" control message. The third packet P3 carrying the ICMP control message "Time Exceeded" is modified to allow looping congestion information as received. It can be seen that in this second embodiment, the modifications to be made to the nodes are also limited.

In a third embodiment, a new message corresponding to a request for providing the congestion information is provided for a control protocol, for example for the ICMP protocol. The first packet PI, and therefore the second packet P2, then carry this new control message. A response to the supply request is also provided to provide congestion information. The third packet P3 then carries this response message to the supply request.

For all of these embodiments, the initiating node 10 causing the transmission of the first packet PI and thus triggering the transmission of the third packet P3 can supervise the good reception of the latter. In addition, the implementation of the method of obtaining does not cause overloading of the communication network, these exchanges of first and third packets being punctual.

The third packet P3 thus carries information relating to the congestion of the route of the node 10 to the node 14 which must not be modified by intermediate nodes carrying the third packet to the node 10.

In a particular embodiment, it is therefore provided that the intermediate node or nodes carrying the third packet P3 in the uplink direction, that is to say from the node 14 to the node 10, verify in the transfer plan whether the third packet P3 carries a control message belonging to the group comprising an ICMP message "Echo Reply", an ICMP message "Time exceeded" or even the response to the supply request. If this is the case, the intermediate node does not modify the information field 31 of the third packet P3, even if it encounters a congestion situation on this upstream direction.

Subsequently, a packet communication network 1 is set up in which the nodes implement differentiated routing of the packets according to classes of CoS services for "Class of Service". This is for example a communication network in Diffserv mode, for "Differenciated Services". In such a network, the packets carry information relating to their class of service. For example, for four classes of service rated from 1 to 4, a service class packet 1 has a higher priority level; a service class 4 packet has the lowest priority level. In such a network, there is provided for the three embodiments a variant allowing a node to obtain information relating to congestion for the routing of packets for a given class of service. Indeed, in such a network, a queue is generally provided for each class of service. A class of service can then be disrupted if the traffic generated by packets of this class is greater than the processing capabilities of the node.

The first PI packet transmitted by the node 10 includes an information field 30 including the class of service for which the congestion information is requested. Thus the node 10 can obtain information relating to the congestion of the given class of service for the route to the node 14.

Optionally, an intermediate node may also set the congestion information for all higher level service classes than the given class of service.

An exemplary implementation of the obtaining method in a communication network in Diffserv mode will now be described for illustrative purposes.

In this example, the node 10 acts as a content server, providing the functions of an audiovisual service platform and integrating a control device. The control device may also be external to the content server.

The content server 10 is able to deliver data packets forming a video stream to client terminals, for example the node 16, through the packet communication network 1 via the nodes 12, 14. The client terminal 16 corresponds for example to a residential gateway, forming the entry point to a residential network of a user. The node 12 corresponds to a router and the node 14 corresponds to a communication network access equipment, for example a DSLAM multiplexer, for "Digital Subscriber Access Line Multiplexer", which is connected to the client terminal 16. The node 18 corresponds to a piece of equipment of the residential network, for example a terminal of the PC type.

It is understood that the invention applies to other contexts than the one just described. For example, the PC type terminal can be connected directly to the network without the intermediary of a gateway. Likewise, node 16 may be the entry point of an enterprise network. Node 14 can be of any type DSLAM as already mentioned, or alternatively MSAN for "MultiService Access Node" in ADSL technology, for "Asymmetric Digital Subscriber Line", or FTTH optics, for "Fiber To The Home" .

The IP communication network in Diffserv mode simultaneously supports priority services called "Premium" and non-priority "Best Effort". As long as the proportion of "Best Effort" versus "Premium" service traffic remains significant in volume, no special mechanism other than the Diffserv prioritization mechanisms is required. Indeed, the simple prioritization of the data flows associated with "Premium" services on data flows associated with "Best Effort" flows ensures that the data flows associated with "Premium" services have a satisfactory level of service quality, possibly by removing some of the associated data flows "Best Effort" services when load rates become very high.

On the other hand, when the proportion of "Premium" services becomes large in volume, it is no longer possible to rely solely on the Diffserv prioritization mechanisms to save the data flows associated with "Premium" services. In this case, only admission control functions limiting the establishment of new communications of "Premium" type and using congested resources, or else rate adaptation of "Premium" type communications, for example to using SVC, make it possible to ensure that "Premium" type communications already established have a satisfactory level of quality.

The implementation of the obtaining method makes it possible to trigger the implementation of these functions.

In step El, the content server 10 inserts ICMP "Echo Request" messages for the "Premium" services to the nodes 14, 16 forming the path to the final client.

On reception (step F1) of a first packet PI carrying the ICMP message "Echo Request", the destination node indicates (step F4) in a third packet P2 carrying an ICMP message "Echo Reply" a possible congestion affecting the downstream direction of the "Premium" services, that is from the content server 10 to the end customer. This is for example a threshold overflow in a queue.

On receipt of this explicit notification, the content server 10 then implements the new communications admission control function for the "Premium" services and using the congested resource. The release of the content server 10 is done after a certain period without receiving congestion notifications. The "Best Effort" services are not affected by the implementation of the process for "Premium" services. Thus, even if a low priority class, for example for "Best Effort" services, reaches saturation, there is no impact on the "Echo Reply" messages transmitted for the higher priority class.

The proposed solution is adapted and particularly suitable for systems in which are defined different classes of traffic with priority of certain classes on others.

Node equipment of the packet communication network will now be described in connection with FIG. 4. Such node equipment notably comprises:

a transmitting and receiving module 102, arranged to transmit and receive packets in the transfer plane;

a module 104 for processing the packets, arranged to process the packets to be routed;

a control module 106, arranged to exchange control information in the control plane with another node equipment; a congestion detection module 108, arranged to detect congestion likely to degrade a routing by this node equipment of packets transmitted by the other node equipment.

The control module 106 is furthermore arranged for:

receive via the module 102 and from the other node equipment a first packet PI,

processing the first PI packet and depending on the latter, requesting the module 102 to send a control message to the other node equipment.

The module 102 is also arranged to process packets in the transfer plan, and in particular to obtain from the first packet PI, to which the control message responds, information relating to congestion of the route in the downstream direction, copy this information in a second packet to be sent. This second packet carries the control message transmitted by the control module 106.

In one embodiment, the control module 106 is also arranged to trigger the sending of the first packet PI to any of the nodes of the communication network 1.

As previously described, the first PI packet may comprise an ICMP control message as described in connection with the first and third embodiments.

The first PI packet may also be a packet for which the validity period expires, as described in connection with the second embodiment. In this case, the packet processing module 104 detects during the processing of the packet that the lifetime has expired and then transmits this information to the control module 106.

In a particular embodiment, the processing module 104 is also arranged not to modify the heading of second packets, when they carry a particular control message, as described above.

The control module 106 of the node equipment is arranged to implement those steps of the previously described obtaining method executed by the node equipment. These are preferably software modules comprising software instructions for executing those of the steps of the method of obtaining described above, implemented by a node equipment. The invention therefore also relates to:

a program for node equipment, comprising program instructions intended to control the execution of those of the steps of the method of obtaining described above which are executed by said node equipment, when said program is executed by a processor thereof;

- a recording medium readable by a node equipment on which is recorded the program for node equipment. The software modules can be stored in or transmitted by a data carrier. This may be a hardware storage medium, for example a CD-ROM, a magnetic diskette or a hard disk, or a transmission medium such as an electrical signal, optical or radio, or a telecommunications network.

The invention also relates to a communication system in a packet communication network, wherein first and second nodes exchange control information in a control plane. The system includes:

a first node, comprising a control module 106 arranged to send to a second node a first packet, for which the second node must transmit in response a control message;

a second node as described above.

The preceding examples are only possible modes of implementation of the invention which is not limited thereto. In particular, other networks may be involved, such as ATM networks.

Claims

A method of obtaining by a first node (10) congestion information of a route for routing packets from said first node to a second node (14-18) in a network (1). ) packet communication, said congestion being capable of degrading said routing, said obtaining method comprising the following steps implemented by the second node:

a step of receiving (F1) a first packet from the first node and destined for the second node, conveyed via at least one third node (12-16), said first packet being designed to trigger issuing a control message in a control plane in which the first and second nodes exchange control messages;

- A processing step (F2) of said first packet in the control plane, after which a transmission of the control message to the first node is requested;

a step of obtaining (F3) said information from the first received packet;

a step of sending (F4) a second packet to the first node, said second packet carrying said control message and comprising said information obtained.

The method of obtaining according to claim 1, wherein the first and second nodes implementing differentiated routing of packets according to classes of service, the first received packet comprises information relating to a given class of service and said information. obtained relates to congestion affecting at least said class of service.

3. Method of obtaining according to claim 1, wherein the control message sent corresponds to a response message to another control message received in the first packet.

The method of obtaining according to claim 3, wherein said other control message is an echo request transmitted from the first node to the second node.

The obtaining method according to claim 1, wherein the header of the first packet indicates that a lifetime of said first packet has expired.

6. Method of obtaining according to claim 1, wherein the second packet being transmitted to the first node through at least a fourth node, said fourth node does not change the header of the second packet.

7. Node in a packet communication network (1) arranged to exchange control information in a control plane with another node:

receiving means (102), arranged to receive from the other node a first packet, said first packet being provided to trigger the transmission of a control message in the control plane;

- Processing means (104) of said first packet in the control plane, arranged to request transmission of the control message to the other node after the processing of the first packet;

means for obtaining (102) information relating to a congestion of a route allowing routing of packets from the other node and to said node from the first packet;

A communication system in a packet communication network, wherein first and second nodes exchange control information in a control plane, said system comprising:

said first node, comprising sending means (106) arranged to send a first packet, said first packet being designed to trigger the transmission by the second node of a control message in response in the control plane;

said second node being arranged according to claim 7.

9. Computer program comprising instructions for implementing the method of obtaining according to claim 1, implemented by a node of a packet communication network, when this program is executed by a processor.